Don't use CFD before understanding these 3 key points

I felt a deep responsibility to write this article because I think everyone deserves benefiting from the power of CFD in the most efficient and effective way. Simply because it's a magnificent tool that is vastly underused. Every process engineer needs some basic education on the tool. And I sensed that this education is largely missing, because I observe too many people struggling with questions such as 'how do we know it can help us?', 'who understands our problem?', 'how can we ever judge if it's done right?' 'should we do this in house or outsource?'

Computational Fluid Dynamics (CFD) is a mature and extremely powerful troubleshooting, design and analyisis tool for modern days. Simply because the models are available and computers have never been so powerful. The barrier to use it in proportion to the value it brings is now a very interesting economic ratio. Managers worldwide are now also more open to embrace it compared to let's say 10 years ago. And this makes this article especially relevant.

This is what you want to get out of CFD

You want to reach your goals timely, at predictable cost. That's really the core of it. Your goals can either be to:

1. Optimise a process by solving an operational issue or saving operational costs
2. Design a process effectively for optimal performance and minimal capital and operational cost
3. Scale-up or scale-down a process while keeping testing time and cost minimal
4. Analyse a situation (e.g. perform a risk analysis to convince regulators)

The 3 key points you should really understand

CFD is a hard core specialist craft. It is no different from a pilot or a surgeon: these specialists need years of training + experience ('flying hours') and their decisions impact you enormously. They should, actually... why else are you doing CFD? I guess you want such experts to know what they are doing. CFD can impact million dollar investments that are built to last for the next 20-30 years. The following 3 key points should be the core of this discussion:

If these criteria are met (where the 3 circles overlap),your CFD project will be succesful and of high value. And your company management will agree.

Key point 1: understand what quality is

With present-day commercial CFD software, anyone of us can produce some nice colorful drawings. But the real specialist can produce the ones that are accurate and truly useful. Quality is the result of two factors:

1. Expertise: the person conducting the CFD work is well trained and has a sound fundamental CFD basis, knows the impact of a certain model selection or setting. Making the right geometry, mesh (the small cells),selecting the right models, picking the right boundary conditions, ... it all requires different sub-areas of expertise. For example: for a person that succesfully finished a '1-phase' project (e.g. water only),starting a 2-phase project (e.g. water + bubbles) is not at all easy... . Hopefully your surgeon knows how to handle the right equipment.

2. Experience: a) preferably, the expert has succesfully completed similar cases in the past. It means, this type of case and/or problem sounds familiar, and the expert figured out how to solve it before. Most of this past experience can then be used. b) the expert should not disconnected from the industry or problem. If this is the case, quite some effort will go to getting the disconnected CFD-expert understanding the real problem. Some fancy simulations will not help if the real problem to solve is not understood. This is 'industry experience'. For example, it's just great if the modeller knows what 'COD', 'DOC', 'aeration', means, and if he/she's familiar with the main hydraulich challenges. c) experience also tells the expert how to process and present the results in a way that maximises the value of the project. Some postprocessing methods can really make the difference. And some methods get top level management super excited... But the expert should know this.

Key point 2: don't forget that the expert should be available to help you

Now it gets interesting. We actually observe 4 main categories of organisations and companies (technology companies, SMEs and large industrial and municipal end users):

1. Group 1: organisations that tried to build in-house CFD expertise and now get rid of their software license and start outsourcing to specialised partners. Some stories are painful
2. Group 2: organisations that never used CFD but now start outsourcing to specialised partners
3. Group 3: organisations that never used CFD but are building or planning to build in-house CFD expertise and buy one or more commercial software licenses and hire one or more CFD experts
4. Group 4: organisations that are not using CFD at all and don't plan to do so yet

The main hypothesis of organisations seeking to build in-house expertise is that a) it will be more cost effective and b) expertise will be readily available. We have however seen so many cases where point a) was never assessed and/or point b) did not work out. And this makes organisations shift from one group to another. What follows is really important if we discuss in-house vs outsourcing:

1. Realise that commercial CFD software and hardware is expensive IF you don't use it almost every day to solve practical problems. It's like buying a professional camera while you're not even a hobby photographer. So you really need a lot of CFD work to be done. One license typically is around 25-30K$/year. The cost of a CFD engineer is much higher. You will easily exceed 100K$/year for just one expert.
2. A CFD expert should be doing it almost full time. Surgery is also not a hobby, and pilots need to fly regularly. Do you have the work for at least one full time person?
3. But one CFD expert is on an island. Even if this person has enough work, he has no critical mass. His learning will be severely hampered, and who whithin the organisation will understand him and recognise his contribution? And did you assess the risk of that person leaving, (in some cases to the competition)? And with that person, the expertise and experience (key point 1) also leaves. No, it's not an exception or a worst-case scenario. We see it happening all the time...
4. In key point 1 we discussed that the expert needs to be able to judge on what he is capable of doing comfortably. a) is this person able to make this judgement? Because no one else in the organisation can. b) If this person can, does this person have the humility to admit that additional expertise is needed? For example your doctor sending you to a neurologist. c) If this person meets a and b, does he dare to say so to his management? Will management support this? Do you have the culture enabling your expensive expert asking for additional expertise? Will you understand?
5. We have many customers with in house CFD expertise. Why? Very often, the in-house experts are working in another team, on different priorities. It's often more difficult to reach them than a professional CFD service partner. Or second, the in-house experts realise and bring up that what is asked now needs dedicated experts. And professional outsourcing typically leads to a better result, faster.

Don't get me wrong, I'm not saying that outsourcing is bad. I'm just saying that the barrier to do so is lower, and that in the vast majority of the cases, it's just a wise decision. Ask yourself what your core business is or should be... The 21st century is the age of outsourcing, because everything has become so specialised. We at AM-TEAM outsource all the time to be able to focus on our core activity: CFD modelling.

Key point 3: realise that CFD can go fast, and you don't have to wait long

We already discussed experience. An experienced CFD person is capable of making objectives crystal clear and knows which strategy to take to meet them fast. Experienced professionals outcompete non experienced professionals with orders of magnitude. Some people still have the perception that it takes months to finish CFD projects. We have finished projects within a week, and typically they all finish between two and eight weeks... Due to bad experiences and strategies, some people tend to conclude that 'cfd takes long'. I hope this article allows you to question your expert and strategy.

Conclusions

Remember, think about these 3 key points, and ask the right questions to assess each of them. And if you then decide to outsource, outsource to a partner that can prove he meets all 3 key points clearly, combined with extreme transparency on cost and deliverables. Remember: you want to reach your goals timely, at predictable cost.

If you have read this article carefully, it will normally help you in your strategic thinking. I have really written this with your interest in mind, and I'm sure that many people in our industry are not aware of these points. Our mission is just to maximise the positive impact of advanced modelling on society. And informing people is part of that mission. If people use more CFD because of this, we succeeded.

Do you have any feedback, or can you tag a person that can benefit from reading this? Curious to hear your opinion.

Wim Audenaert, CEO AM-TEAM (LinkedIn article version)